Project Tango

» see developers.google.com/tango as it lived, on the wayback machine

~ the last words ~

project tango is the kind of project i still pay close attention to. depth-aware phones with onboard slam. exactly the stack i now build with on drones, ten years later. the difference is that tango shipped to consumers in 2016 and almost nobody had a use for it.

google atap announced tango on 5 june 2014 at i/o. the team was led by johnny lee, who had worked on kinect at microsoft before joining google. the early hardware was internal: a 5-inch developer phone codenamed peanut, then a 7-inch yellowstone tablet. both shipped to developers only.

the first consumer device was the lenovo phab 2 pro in november 2016. a 6.4-inch phone with a depth camera plus a fisheye motion-tracking camera. the rgb sensor sat next to them. the asus zenfone ar followed in 2017. that was the commercial inventory at end of life.

~ the record ~

~ what tango actually did ~

area learning let the phone remember a room and re-localize when you came back. motion tracking ran without gps, dead-reckoning through inertial plus visual cues. depth perception measured distance to surfaces with structured light. ar apps would build a 3d mesh of your living room and place a virtual coffee mug on the actual coffee table. it worked when it worked.

the problem was that nobody had built an app worth carrying a thicker phone for. the lenovo phab 2 pro was a brick and the camera array drained battery. the use cases google demoed (indoor navigation, furniture preview) needed buy-in from app developers who could not see a market of more than two phones.

~ how it ended ~

google announced arcore in august 2017. arcore did slam and motion tracking with software only, no depth sensor required, on any modern android. that killed the hardware-led pitch overnight. on 1 march 2018 google ended tango support and pointed developers at arcore.

the talent did not vanish. johnny lee stayed at google, and the slam research bled into arcore, then into pixel cameras and street view. the depth-sensor idea came back as the lidar scanner on the iphone 12 pro in 2020, doing roughly what tango had done four years earlier.

~ the legacy ~

tango sits on the wrong side of a familiar pattern: ship the right idea on the wrong hardware budget. the engineers were right about the future. they just shipped two phones into a market that wanted phones to get thinner.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »